An elcap can be reverse charged unless the voltage exceeds about -2V. This is because the aluminium cathode foil also has a "natural" oxid layer due to the unavoidable oxidation of aluminium in the air, which corresponds to a "natural" formation voltage of about 2V for elcaps. So, a reverse voltage up to -2V doesn't matter.

DC case:

If you measure the leakage current of an elcap by increasing the reverse voltage from 0V on, you will find it rather high and becoming higher and higher the closer you come to -2V. The leakage current of a reverse charged elcap is considerably higher than the leakage current of a non reverse charged elcap. So, when charging two back to back connected elcaps, the two caps form a voltage divider, where the reverse charged cap is considerably lower impedant. So, only a small voltage drops across the reverse charged elcap. This mechanism keeps the reverse voltage well below the -2V limit.

AC situation:

Non-polar or back to back connected elcaps are often used in audio highpass filters. Here they also form a (frequency dependent) voltage divider in combination with a suited resistor. In usual RC-highpass filters the voltage that drops across the back to back connected elcaps does never reach 2V, even at the lowest frequencies and the highest amplitudes. So, none of the both is ever in danger to reach the -2V limit. But even if so, the same would take place as in the DC situation above.

The above is only true if the two back to back connected caps are not very different. So, it's wise to prefer the specially fabricated non-polar elcaps over simple back to back solutions.

Kai Klaas